Added State and Zipcode classes

.gitignore

@@ -2,3 +2,5 @@
 *.gch
 exedisambig
 txt2sqlite3
+.fuse*
+*.gdb

DisambigCustomizedDefs.cpp

@@ -38,6 +38,8 @@ template <> const string cAttribute_Basic<cLongitude >::attrib_group = "Patent";
 
 
 template <> const string cAttribute_Basic<cStreet>::class_name = "Street";
+template <> const string cAttribute_Basic<cState>::class_name = "State";
+template <> const string cAttribute_Basic<cZipcode>::class_name = "Zipcode";
 
 
 template <> const string cAttribute_Basic<cCountry>::class_name = "Country";

DisambigCustomizedDefs.h

@@ -124,6 +124,18 @@ class cStreet: public cAttribute_Single_Mode <cStreet> {
 
 };
 
+// Modeled after the Street class
+class cState: public cAttribute_Single_Mode <cState> {
+public:
+	cState(const char * source = NULL) {}
+};
+
+// Modeled after the Street class
+class cZipcode: public cAttribute_Single_Mode <cZipcode> {
+public:
+	cZipcode(const char * source = NULL) {}
+};
+
 class cCountry: public cAttribute_Single_Mode <cCountry> {
 public:
 	static unsigned int const max_value = 2;

DisambigEngine.cpp

@@ -2034,8 +2034,6 @@ bool fetch_records_from_txt(list <cRecord> & source, const char * txt_file, cons
 
                 std::cout << "Here3..." << std::endl;
 
-                if (i > 4) break;
-
 		if ( pointer_array[i]->get_attrib_group() != string("None") ) {
 			++position_in_ratios;
                 }
@@ -2159,6 +2157,12 @@ cAttribute * create_attribute_instance ( const string & id ) {
 	else if ( id == cStreet::static_get_class_name() ) {
 		p = new cStreet;
 	}
+	else if ( id == cState::static_get_class_name() ) {
+		p = new cState;
+	}
+	else if ( id == cZipcode::static_get_class_name() ) {
+		p = new cZipcode;
+	}
 	else if ( id == cCountry::static_get_class_name() ) {
 		p = new cCountry;
 	}

config/engine.txt

@@ -5,7 +5,7 @@ NUMBER OF THREADS = 2
 GENERATE STABLE TRAINING SETS = true
 USE AVAILABLE RATIOS DATABASE = false
 THRESHOLDS = 0.99, 0.95
-NECESSARY ATTRIBUTES = Firstname, Lastname, Street, City, State, Country, Zipcode, Lat, Lon, InvSeq, Patent, AppDateStr, Assignee, AsgNum, Class, coauths, invnum_N
+NECESSARY ATTRIBUTES = Firstname, Lastname, Street, City, State, Country, Zipcode, Latitude, Longitude, InvSeq, Patent, AppDateStr, Assignee, AsgNum, Class, coauths, invnum_N
 ADJUST PRIOR BY FREQUENCY = true
 DEBUG MODE = false
 NUMBER OF TRAINING PAIRS = 1000000

data/invpat.csv

@@ -1,4 +1,4 @@
-Firstname,Lastname,Street,City,State,Country,Zipcode,Lat,Lon,InvSeq,Patent,AppDateStr,Assignee,AsgNum,Class,coauths,invnum_N
+Firstname,Lastname,Street,City,State,Country,Zipcode,Latitude,Longitude,InvSeq,Patent,AppDateStr,Assignee,AsgNum,Class,Coauthor,invnum_N
 AARON J,TELLER,,WESTBOROUGH,MA,US,1582,4217,-7161,1,03859417,19710513/20020228,TELLER ENVIRONMENTAL SYSTEMS INC,10371134,,D.ROY-0/S.LIN-0/J.LAGARENNE-0,03859417-1
 ABRAHAM,NUDELMAN,,REHOVOT,PA,IL,19004,3103,3456,1,03860581,19720204/20071227,YEDA RESEARCH AND DEVELOPMENT CO LTD,,514/544/564,A.REPHAELI-0/A.PATCHORNIK-0/R.MCCAULLY-0/I.GILAD-0/S.BELL-0/A.WEIZMAN-0/E.KEINAN-0/J.HERZIG-0/M.SHAKLAI-0/F.HAVIV-0,03860581-1
 ABRAHAM A,DOR,,LAKEWOOD,OH,US,44107,4148,-8179,2,03860696,19710325/19751215,HANNA MINING COMPANY,,,P.MCGAULEY-0,03860696-2
@@ -1140,4 +1140,4 @@ ACHYUT,KANTAWALA,5 HASKELL ST APT 41,CAMBRIDGE,MA,US,2238,4237,-7111,1,D0510823,
 AARON,CHU,,TAI-CHUNG,,TW,,2414,12068,0,D0513063,20040409/20060720,GLOBE UNION INDUSTRIAL CORP,,D23/D8/D6,K.HSUI-0/J.LAI-0,D0513063-0
 AARON RICHARD,SERVIER,,SAN FRANCISCO,CA,US,94199,3777,-12241,1,D0513863,20041104,WARNERLAMBERT COMPANY LLC,10176720,D3/D9,R.ARKO-0/S.WAHL-0,D0513863-1
 ADAM P,MALACHOWSKI,,LYNNWOOD,WA,US,98087,4787,-12227,4,D0515012,20041228/20050816,THE BOEING COMPANY,,D12,M.DRAKE-0/P.MIKULENCAK-0/R.EMERY-0/C.NELSON-0/V.STUHR-0/R.WONG-0/C.VEGTER-0/S.BROWN-0/M.EATKINS-0/M.EAKINS-0,D0515012-4
-ADAM L,HILTUNEN,,CHESTERFIELD,MI,US,480
+ADAM L,HILTUNEN,,CHESTERFIELD,MI,US,480